Leadership-Driven Adoption Strategy

Gilbert + Tobin successfully scaled AI by positioning it as a tool for augmenting human judgment rather than a replacement for professional expertise. CEO Sam Nickless spearheaded this by demonstrating his own use of ChatGPT, effectively framing AI as a legitimate productivity tool rather than a shortcut. To ensure firm-wide adoption, the business transformation team moved beyond generic training, embedding themselves in specific departments—such as marketing, finance, and recruitment—to build and demonstrate tailored workflows. This approach resulted in an 87% active usage rate among enabled users, significantly higher than typical tool adoption at the firm.

Governance and Trust

Operating in a high-stakes legal environment, the firm prioritized data security and professional accountability. The rollout was governed by clear policies regarding approved tasks, data entry, and output verification. By leveraging OpenAI’s Australian data residency, the firm met strict internal and client-driven compliance requirements. A core tenet of their strategy is that while AI handles the operational heavy lifting, human professionals remain responsible for constraining the task, reviewing outputs, and applying final judgment.

From Assistance to Workflow Execution

Beyond standard chat interfaces, the firm uses Codex to automate complex, multi-step operational workflows. Notable efficiency gains include:

  • Compliance Checks: Reducing conflict, KYC, and AML checks from 3–8 hours to 5 minutes.
  • Reporting: Automating audit report preparation for 300 entities, saving a full day of manual work.
  • Recruitment: Cutting research and data-extraction time from 4 hours to 20 minutes.
  • Custom Tooling: Developing a 'digital twin' of the CEO to pressure-test ideas and building an AWS monitoring 'watchtower' for the DevOps team.

These implementations demonstrate a shift from discrete task assistance to the automation of irregular, complex processes that are otherwise difficult to justify through traditional software development.